The watermark embedding at low frequencies causes more distortion or less visual quality in the image compared to the watermark embedding at high frequencies. On the other hand, if a watermark is embedded at low frequencies, it can be extracted with a lower bit error rate or greater robustness than if the same watermark is embedded at high frequencies. In this paper, watermark bits are inserted in the DCT high frequency components using the proposed adjusted mean adaptive threshold method. To achieve both robustness and visual quality in the watermarked image, we introduce a parameter, denoted as b, which allows us to adjust the mean adaptive threshold value. The selection of this parameter is crucial as it determines the optimal balance between robustness and visual quality. Through simulation experiments, the appropriate value of b is determined for different attacks such as JPEG quantization, Gaussian noise, and averaging filter applied to high frequency watermarking. By using the minimum appropriate values of "b" equal to 0.5, 2.5, and 2 for JPEG quantization, Gaussian noise, and averaging filter attacks, respectively, the high frequency watermarking at coordinates (7,8) and (8,7) exhibit better visual quality and robustness compared to the low frequency watermarking of reference method, in general.
The rapid spread of misinformation highlights the need for automated fact-checking systems, where claim detection serves as a key initial step. However, developing such systems for low-resource languages like Arabic remains challenging due to limited annotated data and computational constraints. This paper proposes a three-stage approach for Arabic claim detection using the Arabic News Stance (ANS) Corpus. First, we evaluate small, medium, and large Large Language Models (LLMs) in zero-shot settings to establish baselines. Then, parameter-efficient fine-tuning with QLoRA is applied to smaller models. Finally, a lottery-inspired adaptive sampling method enhances supervised fine-tuning by focusing on informative samples and preventing overfitting. The best-performing model, LLaMA-3.1-8B with lottery sampling, achieved 74.78% accuracy, surpassing previous methods with lower resource costs. This work demonstrates that medium-scale LLMs with efficient fine-tuning can provide practical, scalable solutions for claim detection in Arabic.
Despite deep learning approaches capabilities in classification problems, their performances are negatively impacted by insufficient training data, which is most often the case in medical imaging diagnosis. As a result, medical imaging diagnoses using hand-crafted feature-based approaches are still very relevant and efficient, as a stand-alone method and also as a complement to deep learning-based ones, due to generating explorable features. In this research, we devise a new metric (i.e., feature) to classify the pulmonary nodules into benign or malignant. Since benign nodules grow in a more consistent pattern than malignant ones, we hypothesize that the appearance of a benign nodule on consecutive CT image slices is highly correlated. Thus, we propose a linear-prediction-based metric, Linear Nodule Growth Pattern Prediction (LNGP), to capture and quantify any probable inter- and intra-slice correlation in expert-assisted localized nodules, toward malignancy classification. Using the LIDC-IDRI dataset, the results of malignancy classification based on the LNGP metric show AUCs of 0.9998, 0.9478, and 0.9501 over three sub-datasets with increasing uncertainty level, respectively. Being a hand-crafted approach, LNGP-based classification requires less training data for adequate accuracy. Moreover, we show that combining the LNGP metric with conventional hand-crafted methods improves classification performance.
Nowadays, retweeting behavior is considered a substantial behavior to analyze information dissemination in the microblogging network. To understand the underlying mechanism of information dissemination and analyze the evolution of the dissemination corresponding to a given original microblog, user retweeting behavior prediction has recently gained considerable attention. In this paper, we first introduce the four key influential factors that affect the prediction performance of user retweeting behavior then we propose a new measurement based on the random walk for the social influence measurement. Secondly, our defined influential factors are fed into a logistic regression classifier model to estimate the activation of any potential user at each time point. Next, we generalize our model so that, by assuming that we have some activated users concerning the dissemination of a given original microblog at each time point, the future activated user set concerning the dissemination of a given original microblog can be predicted at the next time point. Unlike other comparative models, our prediction model can illustrate the evolution of dissemination and predict the retweeting behavior of all potential users through the dissemination of any original microblog in the microblogging network. Finally, based on the real Twitter microblogging dataset, the experimental results illustrate by the contribution of our defined influential factors and presenting an efficient measurement of the social influence factor, we can simultaneously ensure both the generalization of the proposed model and obtain more accurate predictions in comparison to other baseline methods.
In the future, Internet architectures and decentralized systems will be reshaped by Named Data Networks (NDN) and blockchain technology. Blockchain offers immutability, transparency, and trustless consensus, while NDN emphasizes content over location to revolutionize the networking paradigm. By integrating blockchain and NDN, a secure, decentralized, and content-centric Internet architecture is built. Furthermore, multicasting and caching features of NDN will make blockchain more efficient, while blockchain's decentralized trust mechanisms will increase its security. In this paper, we examine the benefits of combining these technologies, including enhanced trust, scalability, and data provenance. As we explore architectural models, use cases, and open research challenges, we investigate new areas of research. Specifically, we explore two integration paradigms: (1) Blockchain based on NDN, which improves scalability and latency, and (2) NDN enhanced by Blockchain, which secures data provenance and authentication. We assess the benefits and challenges of state-of-the-art solutions. Lastly, we suggest future directions.
Abstract— Today, organizations, especially governmental organizations, are obliged to utilize emerging technologies to survive and respond to rapid environmental changes. One of the most important of these technologies is blockchain, which can play an effective role in enhancing e-government services by creating transparency, reducing corruption, and strengthening public trust. Despite its potential benefits, the acceptance and adoption of this technology in the e-government domain still face various challenges. The present study was conducted with the aim of identifying and analyzing the key success factors in the acceptance of blockchain technology in e-government services. To this end, first, by reviewing the literature and previous studies, the indicators and components affecting the acceptance of this technology were identified. Then, to explain the causal relationships among these factors, the method of fuzzy cognitive maps (FCM) was employed. Based on the developed FCM, scenario-based simulations were conducted to examine the dynamic causal behavior of the blockchain adoption system under different technological, organizational, and institutional conditions. The study population consisted of information technology experts, and field data were collected through interviews and questionnaires and analyzed using SPSS, FC Mapper, and Pajek software. The results led to the presentation of a comprehensive model of blockchain acceptance in e-government, which can be used as a strategic framework for decision-making and policymaking toward the effective implementation of this technology in the public sector.
With the rapid growth and increasing complexity of data, organizations struggle to manage it effectively, thereby revealing the limitations of traditional data architectures. Data Mesh offers a solution through a decentralized, domain-oriented approach that treats data as a product and is supported by federated governance models and self-service infrastructure. Despite its increasing adoption, discussions of its governance within academic circles remain scattered and unstructured. The main objectives of this article are to provide an in-depth overview, share valuable insights into research paths and trends within the field of Data Mesh architecture, and synthesize conclusions from previous studies. This is due to limited domestic research, the fragmented nature of existing studies, and the absence of systematic analysis in this field. Through a comprehensive literature review, this study aims to elucidate key Data Mesh concepts, including distributed data architectures and decentralized governance. By highlighting significant publications, authors, and journals influencing the discourse on Data Mesh, it establishes a benchmark for measuring the breadth and depth of research in this area. One of the aims of this study is to enhance future research trajectories addressing relevant issues. Using VOSviewer software, the methodology integrates bibliometric analysis and systematic review to identify and examine research trends in Data Mesh architecture. Purposive sampling was employed to identify relevant research from the Scopus and WOS databases, with Google Scholar incorporated as a supplementary database for the study's statistical population. The analysis of keyword co-occurrence networks constitutes one of the study’s main findings. Furthermore, four clusters based on citation counts were identified in the co-citation network of sources, with the Computer Science Information Systems and Computer Science Theory and Methods journals forming the largest clusters. A key finding of this research is the identification of significant indicators of Data Mesh architecture through an extensive literature analysis aimed at developing a framework for decentralized data governance. This study provides practitioners and policymakers with evidence-based recommendations for effectively implementing Data Mesh principles. The study also promotes interdisciplinary collaboration by demonstrating the connections between Data Mesh and related domains such as blockchain. Finding future research avenues on the application of decentralized technology in data governance-related concerns is one of the most important results.
This study introduces a novel approach to enhance car insurance fraud detection through the ROME framework, which integrates resampling techniques with cost-sensitive machine learning. The philosophy behind this method stems from addressing two critical challenges in fraud detection: the imbalance in datasets and the high cost associated with misclassifying fraudulent cases. The resampling method ensures balanced data representation, while the cost-sensitive approach prioritizes reducing the misclassification impact, aligning with the industry's goal of minimizing financial losses. This hybrid strategy marks a significant advancement in fraud detection. The model was tested on real-world car insurance data, achieving an impressive F1 Measure of 76.32%, outperforming the CatBoost baseline by 31.25%. These results highlight the effectiveness of the combined approach in enhancing detection accuracy, equipping insurers with a robust tool for improved risk management. The findings offer substantial contributions to the insurance industry by bolstering the reliability and efficiency of fraud detection systems.
3D beamforming is an advanced beamforming method designed to enhance the performance of wireless communication systems. This paper investigates a wireless-powered communication network incorporating an intelligent reflecting surface (IRS) that employs the 3D beamforming method. The IRS consists of numerous passive reflective elements, each capable of adjusting the phase of the reflected signals. Additionally, a UAV serves as a wireless transmitter to communicate with a ground user via the IRS. An optimization problem is formulated to maximize the secrecy rate by jointly optimizing the UAV’s trajectory and the IRS’s phase shift configuration. The energy harvesting (EH) of the user, achieved through the simultaneous wireless information and power transfer (SWIPT) method, is incorporated as a constraint. To solve the optimization problem, convexification techniques, such as Taylor expansion, are employed to alternately optimize the solution. Numerical results demonstrate that the use of IRS technology and 3D beamforming enhances EH and significantly improves the secrecy rate.
As the number of social media users increases, network security becomes more important. The Intrusion Detection System (IDS) is one of the key elements of network security to detect threats and attacks. Since the network data is voluminous and has many features, the feature extraction method alongside classification methods is a common solution to this issue. Convolutional Neural Network (CNN) is a proper feature extraction method to manage the network data complexity. Support-Vector Machine (SVM) is an outstanding method to classify network data moving high-dimensional data to high-dimensional feature space. In this paper, we propose a hybrid anomaly detection system using CNN for feature extraction and Least-Squares SVM (LS-SVM) for classification. The KDDCUP’99 and UNSW-NB15 datasets are considered in our research to simulate real network traffic and cover new attack types. The experimental results show that our hybrid approach improves the F1-score and AUC metrics compared to its counterparts.
This study investigates the behavioral intention of students at the University of Tehran to use ChatGPT as an educational support tool, using an Extended Technology Acceptance Model (ETAM) that builds upon the original TAM framework. Core constructs such as perceived usefulness, ease of use, attitude, and behavioral intention were examined alongside extended variables including privacy, security, trust, social influence, and technical knowledge. Based on data collected from 330 students across various disciplines and analyzed through structural equation modeling, the findings indicate that attitude is by far the strongest predictor of behavioral intention (β = 0.795), followed by perceived usefulness and ease of use. Technical knowledge significantly enhances perceived ease of use but exerts only an indirect influence on attitude and behavioral intention. Although privacy and security concerns remain relevant, their impact is mediated through trust and (to a lesser extent) social influence, and neither trust nor social influence directly affects attitude or behavioral intention. This study confirms the robustness of the TAM framework in explaining AI adoption in education and highlights the importance of perceived usefulness, ease of use, and positive attitude as primary drivers. The findings suggest that universities should focus primarily on improving the perceived usefulness and ease of use of generative AI tools, while continuing to strengthen digital literacy initiatives and address privacy and security concerns in order to support responsible and effective integration of tools such as ChatGPT into teaching and learning.
The advent of Artificial Intelligence (AI) presents both opportunities and challenges for the labor market in developing countries. This paper aims to systematically assess the impact of AI on the labor market in Iran and identify effective strategies to mitigate its challenges. Major impacts of AI on labor market and strategies to properly confront the impacts were extracted using existing literature from 2020 to 2024. Then, a hybrid MCDM approach based on the judgements of a group of technical and policy making experts was implemented to determine which major impacts are more important and what are the best strategies confronting them in the labor market. Similar studies were found in the literature but a study focusing on a developing country like Iran was missing in the literature. The findings of this study offer valuable insights for policymakers, educators, industry leaders, and other stakeholders. By providing a prioritized roadmap of strategies tailored to the specific challenges of AI, this paper contributes to the strategic planning and policy formulation necessary to harness the benefits of AI while mitigating its adverse effects on the labor market. Ultimately, our research aims to guide a balanced and sustainable integration of AI into the workforce, ensuring economic growth and employment stability in the face of rapid technological evolution.
This paper investigates beamforming design for Simultaneous transmission and reflection reconfigurable intelligent surface (Star RIS) assisted secure wireless communication with three operating protocols (energy splitting (ES), mode switching (MS), and time switching (TS)) with the assumption of the availability of imperfect CSI of two eavesdroppers at BS. We maximize the sum rate of legitimate users by designing both active and passive beamforming at BS and RIS, respectively, under the constraint of the limited maximum tolerable rate of eavesdroppers and a limited transmission power budget. The non-convex problem is solved by alternating optimization (AO) for active and passive beamforming optimization, and also successive convex approximation (SCA), and the penalty concave-convex procedure (PCCP). According to the simulation results, Star-RIS outperforms conventional RIS, ES mode outperforms MS and TS, and the performance is higher with lower channel estimation error.
Currency exchange rate forecasting has always been one of the important issues for economic activists. In this context, the stationary, and non-linear behavior of this variable and random walk claim mentioned in some empirical studies have made forecasting as one of the challenges, and concerns in the field of economics. The present study briefly classifies various currency exchange rates forecasting models and methods, then focuses on five deep learning methods, including Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU) and Reinforcement Learning (RL). For this purpose, the report of various studies on forecasting currency exchange rates using the above methods, with objectives, such as identifying the researchers in this field, the scope of studies, the scientific centers conducting studies, along with their geographical distribution, mixed methods used, studied currency pairs, forecasting periods, data frequency, evaluation criteria, obtained accuracy and features used for forecasting were studied. The study results will help future research in this field more effectively identify the research gaps with classified access to the previous studies, and define the topic and scope of future research to complete previous studies.
The popularity of cryptocurrencies has intensified the need for accurate volatility prediction models. This research proposes a novel approach to enhance conditional variance predictions for cryptocurrencies. By leveraging a feature selection technique that selects strong features based on price, return, and volatility cross-correlation analysis, we effectively select the most relevant features for an LSTM-based prediction model. To obtain initial volatility estimates, various GARCH family models (GARCH, EGARCH, and GJR-GARCH) were fitted to the dataset, with the best-fitting model selected based on minimum MSE and RMSE. Subsequently, the proposed MGF-LSTM model was applied to the top eight cryptocurrencies by market capitalization. Experimental results demonstrate that our model significantly reduces prediction errors, providing valuable insights for risk management and investment decision-making in the cryptocurrency market.
Visual Question Answering (VQA) is a complex task that requires models to jointly analyze visual and textual inputs to generate accurate answers. Reasoning and inference are critical for addressing questions that involve relationships, spatial arrangements, and contextual details within an image. In this study, we propose a model based on the BLIP framework, as a generative model, that enhances contextual understanding by incorporating dense-captions -detailed textual descriptions generated for specific regions within an image- along with spatial information extracted from the image. The model focuses on emphasizing visual information and extracting additional context to improve answer accuracy. Experimental results on the GQA dataset demonstrate that the proposed approach achieves competitive performance compared to state-of-the-art methods
In recent years, Convolutional Neural Networks (CNN) have been extensively used in machine learning algorithms related to images due to their exceptional accuracy. The multiplication-accumulation (MAC) in convolutional layers makes them computationally expensive, and these layers account for 90% of the total computation. Several researchers have taken advantage of pruning the weights and activations to overcome high computation bandwidth. These techniques are divided into two categories: 1) unstructured pruning of the weights can achieve heavy pruning, but in the process, it unbalances data access and computation processes. Consequently, compression coding for indexing non-zero data increases, which causes much more memory volume. 2) Structured pruning by the specified pattern prunes the weights and regularizes both computations and memory access but does not support high pruning amounts compared to unstructured pruning. In this paper, we proposed Quasi Structured Pruning (QSP) that profits from the high pruning ratio of unstructured pruning. The load balancing property in structured pruning has also been included in the QSP scheme. Implementation results of our accelerator using VGG16 on a Xilinx XC7Z100 indicate 616.94 GOP/s and 1437.7 GOP/s at just 7.8 watts power consumption for dense and sparse mode, respectively. Experimental results show that the accelerator is 1.38×, 1.1×, 2.77×, 2.87×, 1.91×, and 1.18× better in terms of DSP efficiency than previous accelerators in dense mode. As well, our accelerator has achieved 1.9×, 2.92×, 1.67×, and 1.11× higher DSP efficiency besides 4.52×, 5.31×, 10.38×, and 1.1× better energy efficiency than other state-of-the-art sparse accelerators.
The proliferation of fake news on social networks poses significant challenges for trust, security, and societal well-being. In this paper, we present a comprehensive study of fake news detection approaches and techniques, introducing a novel framework for news construction comprising four elements: news content, news context, news propagation, and news environment. We propose a new taxonomy of fake news detection techniques categorized into two primary types—individual methods (content-based, context-based, and propagation-based) and frameworks (hybrid and perception-aware methods). We highlight their strengths, weaknesses, and applicability by analyzing 14 state-of-the-art detection methods across platforms such as Twitter, Facebook, and Sina-Weibo. Furthermore, we address critical research gaps by identifying future directions, including early fake news detection, unsupervised learning, multimodal datasets, adversarial attacks on algorithms, multi-lingual platforms, and AI-generated content detection. Our findings and recommendations aim to serve as a foundation for developing new robust, scalable, and impactful fake news detection systems.
The rapid adoption of artificial intelligence (AI) technologies across diverse sectors has exposed vulnerabilities, particularly to adversarial attacks designed to deceive AI models by manipulating input data. This paper comprehensively reviews adversarial attacks, categorising them into training-phase and testing-phase types, with testing-phase attacks further divided into white-box and black-box categories. We explore defence mechanisms such as data modification, model enhancement, and auxiliary tools, focusing on the critical need for robust AI security in healthcare and autonomous systems sectors. Additionally, the paper highlights the role of AI in cybersecurity, offering a taxonomy for AI applications in threat detection, vulnerability assessment, and incident response. By analysing current defence strategies and outlining potential research directions, this paper aims to enhance the resilience of AI systems against adversarial threats, thereby strengthening AI's deployment in sensitive applications.
Wireless sensor networks (WSNs) are advanced tools for monitoring and controlling the environment, which are powered by a limited capacity battery, and the depletion of the sensor battery leads to the end of the life of the network life. Therefore, it is crucial to use protocols that are energy efficient. In this paper, using a new cross layer model based on distance from base station (BS) and (Time Division multiple access) TDMA, the optimal use of available resources and increasing of the life time of the network are discussed. By modifying the method of selecting the cluster headers (CHs), the selection of low energy CHs, which are far from the BS, has been prevented. It also balances the transmission of data packets in different clusters, resulting in fair energy consumption between sensors. By comparison with the ATEER model in [12], the proposed model has reduced energy consumption by 45%, increased the life time of the network by 67% and increased packets sent to the BS by 15%.
